Calling X-over designer

ttan98

Member
2006-04-04 11:24 am
Melb
This should be easy to all of you.

I am designing a 2-way speakers. I am thinking of implementing either one of these 2 types of x-over, option A - LR12 or LR24 say at 2Khz, option B - Butterworth filter, BW18 at same freq.

Tell me whether I am right if I am wrong please explain.

Option A

No reversal of polarity of the tweeter is required.

Option B

Reversal of tweeter's polarity is required. Please explain why reversal is required. Is this type of x-over recommended?
This site seems to like 3rd order Butterworth,
http://home.wanadoo.nl/dezaire/R&D.htm

Thanks in advance
 
Reversal or not reversal is given by the actual phase relationships between the drivers around the crossover frequency. This includes:

- the phase difference introduced by (any) filter: 180 deg LR2, 360 deg LR4, etc.
- the phase response of the driver themselves (usually significant phase shift is present in the roll-off regions)
- the phase difference (actually time delay) due to the physical misalignment of the drivers voice coils

ALL these should be considered when connecting the tweeter - failure to do so is the reason why "generic" filters don't work well. In practice, one needs to get phase relationships right for a good speaker.
 

ttan98

Member
2006-04-04 11:24 am
Melb
bzfcocon said:
Reversal or not reversal is given by the actual phase relationships between the drivers around the crossover frequency. This includes:

- the phase difference introduced by (any) filter: 180 deg LR2, 360 deg LR4, etc.
- the phase response of the driver themselves (usually significant phase shift is present in the roll-off regions)
- the phase difference (actually time delay) due to the physical misalignment of the drivers voice coils

ALL these should be considered when connecting the tweeter - failure to do so is the reason why "generic" filters don't work well. In practice, one needs to get phase relationships right for a good speaker.

If the woofer is LR2 and tweeter is LR2 there is STILL a possibility a phase reversal will occur at x-over point due to time allignment, etc.

A very good explanation that is all I need, thanks.
 
Here's a LR2 3-way.

The top graph is with the mid polarity reversed +ve to -ve connection (in phase). The lower graph is with the mid polarity normal, +ve to +ve connection, (out of phase).

LR2 normally has 180° phase difference in the xo which means the tweeter or mid have to be connected in reverse but it doesn't happen all the time with some required normal connection.

What the graph shows that with out of phase there should be deep nulls indicating good phase relationship between the drivers at the xo point. Always a good check when designing an xo.

Pictures worth a 1000 words or something like that.
 

Attachments

  • phase_2.jpg
    phase_2.jpg
    52.5 KB · Views: 203

ttan98

Member
2006-04-04 11:24 am
Melb
rabbitz said:
Here's a LR2 3-way.

The top graph is with the mid polarity reversed +ve to -ve connection (in phase). The lower graph is with the mid polarity normal, +ve to +ve connection, (out of phase).

LR2 normally has 180° phase difference in the xo which means the tweeter or mid have to be connected in reverse but it doesn't happen all the time with some required normal connection.

What the graph shows that with out of phase there should be deep nulls indicating good phase relationship between the drivers at the xo point. Always a good check when designing an xo.

Pictures worth a 1000 words or something like that.


I have seen similar FR graphs before, I just want and explanation.

Thanks for your effort.
 

jnb

Member
2006-12-30 11:55 pm
Originally posted by ttan98 I just want and explanation.

If you look at the phase plot of a crossover, the woofer's phase tilts downwards as frequency increases. The tweeter's tilt upward as the frequency decreases. The tilted regions of the two drivers are parallel to each other but now are separated.

The question is by how much.

In a (text book) second order crossover, the separation is by 180 degrees at all regions within the crossover. It will only cause cancellation where the drivrs are at similar levels to each other. Reversing the tweeter will prevent this cancellation as it changes the relationship by 180 degrees.

In a (text book) fourth order crossover, the phase wraps around 360 degrees and meets up at zero where it started. The two filters meet up with each other (looks like a spiral that starts and ends at zero degrees) and you don't need to reverse one driver, but there still is a lot of phase shift around the crossover.

With a (text book) third order crossover, the phase separation is 270 degrees (-90 degrees). Reversing a driver will make it +90 degrees. There will always be some tilt of the forward lobe.

BTW, a practical text book crossover means the acoustic response can be measured to be just as the book says it should. As drivers do not have flat phase, frequency and impedances, you cannot make such a filter without measuring these parameters in detail and making a custom crossover that produces this result.
 
jnb's explanation is very accurate. If you don't fully understand his explanation, only thing you need to do is to read more elsewhere. This topic can't be fully explained by a few paragraphs in a casual forum post. You'll need to understand what is acoustic phase, how drivers' acoustic rolloffs cause their phase shift across frequencies, and how these affects drivers' amplitude summation. The trasient response of an xover's rolloff rate is another good topic to know. Here's a first good read:

http://www.geocities.com/kreskovs/Phase-B.html
 
ttan98 said:
This should be easy to all of you.

I am designing a 2-way speakers. I am thinking of implementing either one of these 2 types of x-over, option A - LR12 or LR24 say at 2Khz, option B - Butterworth filter, BW18 at same freq.

Tell me whether I am right if I am wrong please explain.

Option A

No reversal of polarity of the tweeter is required.

Option B

Reversal of tweeter's polarity is required. Please explain why reversal is required. Is this type of x-over recommended?
This site seems to like 3rd order Butterworth,
http://home.wanadoo.nl/dezaire/R&D.htm

Thanks in advance
Personally I would not use XOs that required reversing polarity.